Analysing exceptions within qualitative data: promoting analytical diversity to advance knowledge of ageing and physical activity

Abstract Looking for patterns of meaning within data to identify central themes is a common form of analysis within qualitative research in sport, exercise and health. Far less analytical scrutiny has been directed toward how researchers might deal with ‘exceptional’ data. That is, data which, while telling us something about a central theme, deviates significantly from its defining plotline and characteristics. The purpose of this predominantly methodological paper is to examine exceptions in data gathered from interviews with 51 (m = 23; f = 28) physically active older adults (60 – 92 years of age). Exploiting exceptions within our data revealed unique perspectives within the central themes of: healthy ageing, relationships, and bereavement. Problematising the rise of – and indeed pressure for – methodological simplicity within qualitative research, we assert the continued need for complexity for progressing the intellectual agenda of ageing and physical activity. Engaging with methodological multiplicity, particularly at the level of qualitative data analysis (e.g. via a focus on exceptions), produces important and original knowledge, which has direct relevance for the development of theory, methods and health policy.


Introduction
Looking for patterns of meaning within data to identify central themes is a common form of analysis within qualitative research in sport, exercise and health. Notwithstanding important differences between them, thematically driven styles of analysis include thematic analysis (Braun and Clarke 2006), narrative thematic analysis (Riessman 2008), Interpretive Phenomenological Analysis (IPA) (Smith et al. 2009) and framework analysis (Spencer et al. 2014). The prevalence of research examining central themes is, in many respects, unsurprising. After all, Saldaña (2011) explains how the human propensity for pattern making follows us into qualitative data analysis as an almost 'intrinsic, hardwired need to bring order to the collection [of data] -to not just recognise it, but to look for and construct patterns out of it ' (p. 91). The value of this style of analysis is highlighted by Braun and Clarke (2016) who note that focusing on central themes within a data-set can: Similarly, Sparkes and Smith (2014, p. 124) reflect on the straight forward and flexible nature of examining central themes, suggesting that it pushes the researchers 'towards a deep, freewheeling, aesthetically satisfying interpretation of the data' . Green and Thorogood (2004) meanwhile, point out its usefulness for identifying typical responses to questions posed to a group, thereby gaining an awareness of the important issues.
No analysis, however, is ever perfect. Referring specifically to thematic analysis, advocates and innovators of this analytical form, Braun and Clarke (2016), contend that despite its value and popularity, qualitative research does not start and end with thematic analysis. Rather, in some instances, there is a need to move beyond the summaries of participant accounts that a singular focus on central themes can provide (Green and Thorogood 2004). This is because in the process of looking for patterns of meaning to identify central themes, the contradictions and complexities inherent to data can at times, be 'ironed out' (Sparkes & Smith 2014, p.126). One response to this concern, has been to use multiple forms of analyses. Indeed, Coffey and Atkinson (1996) assert that the more we examine our data from different viewpoints, 'the more we may reveal -or indeed construct -their complexity' (p. 14, see also Smith and Sparkes 2006;. Within sport, exercise and health, subjecting qualitative data to multiple forms of analyses is increasingly utilised (see Clarke et al. 2016). This has been particularly the case within narrative inquiry, where using pluralistic data analysis has offered detailed insight into a range of issues. These include: athletes experiences of injury (Phoenix and Howe 2010), positive ageing identities and physical activity (Phoenix and Sparkes 2009), distance runners experiences of disordered eating (Busanich et al. 2012(Busanich et al. , 2014, transition from professional sport (Douglas and Carless 2009), experiences of elite golfers (Douglas and Carless 2006), physical activity rehabilitation among spinal injured populations (Papathomas et al. 2015) and social comparison within cancer experiences of elite athletes .
Amidst this dialogue around qualitative data analyses, one analytical strategy that has been largely overlooked is how researchers respond to and potentially use exceptional data. By this we mean data which, while telling us something about a central theme, deviates significantly from its defining plotline and characteristics (McPherson and Horne 2006). Exceptions, therefore, constitute examples of the very nuances and contradictions within our data that can be so easily missed throughout the analytical process. Within quantitative research, exceptions to the majority are often termed 'outliers' and generally eliminated from statistical analysis due to the significant yet somewhat unjustified influence they might have on statistical results. Within qualitative research, however, far less analytical scrutiny has been directed toward how researchers might deal with exceptional data. This is an important omission, which has implications for what is claimed to be 'seen' within qualitative data and subsequently the rigour and credibility of any concluding statements that might be drawn from it. Overlooking exceptions, therefore, is problematic because it can overshadow the confidence held in one's thematic findings (McPherson and Horne 2006). With these points in mind, the purpose of this methodological paper is to examine exceptions within qualitative data. Such a focus will enable us to expand the diversity of qualitative data analyses in a way that is significant to the development of the intellectual agenda of sport, exercise and health sciences.
To provide a context and show how the inclusion of exceptions within our analyses can extend our understanding of a given topic, we use empirical data from a larger research project that examined the role of physical activity in older age. 1 Using empirical data in this way enables us to illustrate what an analytical concern with exceptions might look like 'in action' . On this occasion, our application of this analytical focus leads to the reporting of original research findings regarding factors that impact upon later life physical activity. However, the analytical principles and subsequent issues pertaining to research evidence that it highlights are equally relevant to other topics within qualitative research in sport, exercise and health.

Physical activity and ageing
Qualitative research in the field of ageing and physical activity has increased dramatically over the last fifteen years. Qualitative contributions can be found across numerous disciplines, (sadly, often failing to acknowledge each other), including sociology of sport and exercise (Tulle 2008a;Pike 2012); leisure sciences (Dionigi 2006), geography (Winters et al. 2015), gerontology (Paulson 2005;Evans and Sleap 2012), health psychology (Ayotte et al. 2010) and public health . It is beyond the scope of this paper to offer a full and detailed account of the knowledge gleaned from this literature. That said, several key issues have been identified as being important in facilitating and hindering physical activity participation in older age. These include: social influences (valuing interaction with peers, encouragement from others); physical limitations (pain or discomfort, concerns about falling, comorbidities); competing priorities; access difficulties (environmental barriers, affordability); personal benefits of physical activity (strength, balance and flexibility, self-confidence, independence, improved health and mental well-being); and motivation and beliefs (apathy, irrelevance and inefficacy, maintaining habits) (see Franco et al. 2015).
The steady growth of qualitative scholarship in this area of ageing and physical activity over the last decade has been illustrative and informative, but has reached an important juncture. The lessons that can be learned from interviews with older people (usually 'young-old' women, rarely rural dwelling, nearly always retired, white, able-bodied, and financially secure) about their experiences of sport and physical activity are diminishing. Doing little more than identifying central themes from such data (often rather crudely expressed as either 'barriers' or 'facilitators' to physical activity) has been insightful in the past, but has less to offer moving forward. More of the same is not enough if we are to progress the intellectual and impact agendas of this field.
One strategy for supporting progress is to engage critically with notions of later life physical activity by connecting the meanings that people ascribe to individual experience with wider social and cultural contexts. Scholarship in this domain (e.g. see Phoenix 2017;Tulle and Phoenix 2015) has discussed the potential of physical cultures to resist dominant narratives of ageing, which pathologise bodily change and reinforce narratives of decline (Dionigi 2006;Tulle 2008b;Phoenix and Smith 2011). It has also highlighted the significance of embodied pleasures (Fullagar 2012;Nettleton 2013;Phoenix and Orr 2014;Jette and Vertinsky 2015), life history (Tulle 2017), and space/place encounters (Bell and Wheeler 2015;Sparkes 2015) that shape experiences of physical activity in older age, and provide important additions to the dominant health promotion messages regarding exercise dose-response relationships (how much, how often, how long, how intense). These latter avenues of inquiry however, remain in their infancy.
Another strategy, is to champion diversity within our analytical repertoire. Indeed, as an increasing body of qualitative studies about physical activity in older age begins to construct and confirm a homogenous impression of it we could easily become misled into the conviction that nothing more can be learned. Under such circumstances, McPherson and Horne (2006) urge that 'the study of important phenomena benefits from the application of a diversity of increasingly sophisticated research methods to ensure that all relevant angles of vision are exploited and all reasonable interpretations carefully considered' (p. 7).
To that end, in this paper, we present three examples of how exception (or 'outlier') data was incorporated into our analyses. Using these examples, we demonstrate how it can provide a useful analytical tool that has the capacity to move our knowledge of later life physical activity beyond 'more of the same' , by illuminating original insights that advance the field in new and thought provoking ways.

Methodology and methods
The broader research project from which this paper draws, was informed by narrative inquiry, and specifically narrative constructionism. Narrative inquiry involves the reconstruction of a person's experience in relationship both to the other and to a social milieu (Clandinin and Connelly 2000, for an overview of narrative inquiry's contribution to qualitative research in sport and exercise see Papathomas 2016). Inspired by Gergen (1999), in their discussion of narrative as a dimension of constructionist inquiry, Sparkes and Smith (2008) use the term narrative constructionism to signal an understanding of narratives being 'a vehicle through which our world, lives and selves are articulated and the way in which such narratives function within social relationships' (p. 298, original emphasis). Thus, from this perspective narrative is seen as a form of social action and a relational sociocultural phenomenon used to configure and constitute experience and make sense of who we are.
The project was structured using two linear yet complementing studies. Study one -which yielded the data upon which our interest in exceptions is based -focused on the meaning and experiences of physical activity for older adults who exercise on a regular basis (see Authors). Study two was concerned with how other people at various stages in the life course respond to such accounts (see Authors). For study one, we used criterion and snowball sampling to identify our participants (Sparkes and Smith 2014). Our inclusion criteria stated that participants needed to be age 60 years or over and self-identify as exercising on a regular basis. We sought diversity within the sample regarding age group, activity type and equal numbers of males and females. After receiving ethical approval, we recruited 51 participants (M = 23; F = 28) from the south west of England, who (it transpired through the data collection), also spanned a range of health conditions including chronic obstructive pulmonary disease, heart disease, arthritis, diabetes, cancer, Parkinson's disease, depression and anxiety. They were recruited using a variety of means including: displaying posters about the project in public spaces (e.g. libraries, leisure centres); approaching naturally occurring interest groups for this age group (e.g. exercise groups, lunch clubs). Those who expressed an interest in the project were provided with information sheets outlining the purpose of the research, the nature of their potential involvement and details of any ethical issues (e.g. confidentiality). Informed consent forms were then circulated and discussed prior to commencing each interview.

Analyses
The data were subjected to multiple forms of narrative analysis. As explained by Smith (2015), 'narrative analysis focuses on one specific genre of discourse, that is, a story. Other qualitative approaches and analyses are interested in all genres of talk and text' (p. 210). We began with a categorical-content analysis (Lieblich et al. 1998), or what Riessman (2008) terms a thematic narrative analysis. This analysis focuses on the whats of talk by examining the core pattern, or key themes in each participant's life story (Sparkes and Smith 2014). It responded to the aims of the project by illustrating a number of different roles that older adults ascribed to participation in physical activity. These included: contributing to healthy ageing, developing and maintaining relationships, enhanced sociability, assisting in the bereavement process through the development of new social networks, providing experiences of pleasure, resisting stigma. Next, our analytical attention turned to the practical organisation and meaning of how these themes were encountered through experiences of physical activity. This involved asking questions in relation to the what and where of social context (Holstein and Gubrium 2004). Asking such collective questions facilitated a means of working across social settings and making visible the ways in which alternative forms of meaning making were being accomplished (see e.g. Authors).
The extensive sifting through the data and ongoing periods of indwelling required by these analyses, sensitised us to occasional fragments of stories that did not 'fit' with the recurring storylines emerging as key themes via our categorical-content analysis. Despite constituting what might be regarded by some as 'outlier data' and subsequently discarded, these fragments quite simply, stayed with us. Or, drawing from Frank (2010, p. 48), they got under our skin. Although distant from the plotlines and characteristics that had come to define our key themes, these stories generated 'intense, focused engagement' , affecting how we thought, knew and perceived those themes. Thus, rather than discard these exceptions as outlier data that had nothing to offer, instead we turned our analytical attention to moving through them more fully.
The notion of movement in relation to narrative analysis is discussed by Frank (2010), who like many other narrative scholars (e.g. Riessman 2008; Sparkes and Smith 2012) bemoans the dependency on highly prescriptive analytical guidelines, which, he believes prevents interpretive thought from moving. For Frank, 'Analytic or interpretive thought that is moving is more likely to allow and recognise movement in the thought being interpreted' (p. 73). Smith (2015) also encourages movement of thought throughout the analytical process, asserting how 'it can take the analyst in unexpected and fertile directions, breathing fresh life into moribund concepts, encouraging theoretical curiosity and provoking new ways of seeing in the process ' (p. 213-214). Thus, as we turned our attention to exceptions within our data, our movement of thought involved asking searching questions of the exceptions we had identified rather than following a predetermined prescriptive procedure that might limit or artificially neaten the depth and nuance that we were aspiring to highlight. These questions resonated with those informing a dialogical narrative analysis and included: resource questions (what resources does the storyteller utilise to shape their experiences?), circulation questions (who would immediately understand this story and who would not understand it?), connection questions (to whom does the participant's story connect them?), identity questions (how does this story give its teller a sense of who they are?), body questions (what is our body telling us about the story, the storyteller and what it means to live well?), function questions (how might this story be good for the teller? How might it harm them?) (Frank 2010;Smith 2015).

Exceptional data in later life physical activity: three examples in action
In what follows, we describe and discuss examples of exceptions identified within three of the central themes that were generated through our categorical-content analysis: healthy ageing; interpersonal relationships; and death and bereavement. Our intention in exploiting exceptions was not to replace the central themes, but to forge new ground by embracing complexity within them through explicitly focusing on the contradictions and challenges within accounts of physical activity in older age, that can otherwise seem very familiar. To be clear, in talking of exceptions, we are not referring here to exceptional individuals within our sample. To focus on exceptions within the analysis is not to focus on heroic narratives of extraordinary people or achievements (the problems of which have been outlined elsewhere, see Author, ****). Rather, it is to attend to those occasional fragments of stories that do not 'fit' with the recurring storylines emerging as key themes via a categorical-content analysis and as a consequence, run the risk of being discarded or overlooked.

Healthy ageing and the ambivalent health strategy
One of the central themes in our data reflected the perceived central role physical activity played in ageing with good health. Physical activity was a key component within a broader repertoire of behaviours believed to support healthy ageing, which also included eating well and keeping busy (see Katz 2000). These explanations often connected with broader social narratives pertaining to healthism (whereby good health is understood as an individual responsibility), and the notion of anti-ageing (in which formal exercise becomes a viable weapon in the fight against the ageing process [Tulle 2008b]). Participants largely incorporated the adage of 'use it or lose it' into their stories of their ageing body, reifying Jones and Higgs' (2010) observations regarding the centrality of fitness in what currently constitutes normative ageing.
However, in working with the data, it became evident that there were also significant variations to this theme that proved instructive to the analysis. One exception, while still providing insight into the dominant theme of 'healthy ageing' and contributing to knowledge around what it means to be physically active in older age was the use of physical activity alongside health harming behaviours (e.g. heavy drinking 2 and consumption of high cholesterol foods). These were not stories of curtailment nor planned ceasation, but of co-existance. In these instances, being physically activity was integral to a more ambiguous health strategy, wherein a complex web of health enhancing and typically health-harming behaviours intertwined to redefine notions of ageing well. An example of this exception to the more typical storylines encountered in relation to 'healthy ageing' within our study, is presented using John's account of why it was important for him to continue with his physically active lifestyle: I am doing half marathons now… if I cut my drinking by half and train twice as much, if I got down to 12 stone I would probably have a crack at another marathon. But, I am normally about 13 and a bit stone. I run so I can enjoy the rest of my lifestyle really. I enjoy the running, and I enjoy the beer at my local pub and a glass of wine with supper and maybe a glass of port with some nice cheese and grapes and stuff of an evening. So if you asked my doctor, he will say, come on, you ought to consider changing your lifestyle' , because I admit to about 50 units a week…. So they keep talking about this change in lifestyle. The short answer is, if I didn't do 20 miles a week, if I didn't go biking, if I didn't play a bit of tennis and golf, I would probably be about 17 stone, high cholesterol, high blood pressure, dead ringer for a heart attack or a stroke or both before I was 80. It allows me to live 'normally' as I call it. If I wasn't able, if something happened and I wasn't able to run I would have to seriously think about cutting down my alcohol. (John,age 73) Exploiting the exceptions within our analysis highlighted attempts to balance different priorities such as the experiences of pleasure that can be gained from practices that can harm physical health (e.g. heavy consumption of alcohol and high cholesterol foods) alongside those advocated to maintain or improve health (e.g. regular physical activity). These balancing acts were a key feature in the maintenance of and ongoing adjustment to physical activity regimes. The use of the balance metaphor complies with previous research highlighting how health-relevant decisions are made in the context of daily life, amidst a plethora of competing priorities (e.g. Backett 1992;Pajari et al. 2006). Yet to date, it has rarely been considered in relation to broader accounts of the ageing process and aspirations to age well. Exceptions to this theme of 'healthy ageing' demonstrate attempts to create coherence across conflicting health practices. In practical terms, and reflecting Lupton's (2016) description of the quantified self, this involved the self-tracking and self-monitoring of exercise performed and balancing those against the units of alcohol said to be consumed. Quantification acted as a reconciliation device between two apparently contrasting storylines because it provided a site for different and competing health repertories and their moral components to co-exist. Will and Weiner (2014) extend the notion of reconciliation in their research on everyday cholesterol reduction, repertoires and practices. For them, while talk about finding a balance in people's descriptions of healthy living demonstrate attempts at reconciliation, attempts are also often made to simply accommodate contradictions and live with incoherence. Our exception supports this by showing how quantification supported reconciliation ('if I didn't do 20 miles a week … I would probably be about 17 stone'), while the apparent normalisation of contradictory, yet coexisting health practices facilitated an ability and willingness to live with incoherence. Importantly, the exception enables Will and Weiner's theorising to be extended by connecting it with notions of masculinity. Specifically, we might argue that this strategy (attempts to both reconcile and accommodate contradictorary health practices) also reinforces gendered representations of health and masculinity by minimising the risk of John being seen to take himself and his health too seriously. In this regard, Courtenay (2000Courtenay ( , p. 1397) notes how by dismissing health needs and taking risks, men legitimise themselves as the 'stronger' sex. In this way, men's use of unhealthy beliefs and behaviours help to sustain and reproduce social inequality and the social structures that, in turn, reinforce and reward men's poor health habits.
While these observations by Courtenay are certainly relevant, so too is more recent critical scholarship illuminating a previously overlooked and gendered narrative regarding the need for men to care about doing health work, but not too much (Smith 2013), and the dynamics inherent to maximising masculine capital in relation to men's health (de Visser, and McDonnell 2013;Gough 2013). Far from dismissing health needs in their entirety, this exception signals how the masculine capital accrued (or maintatined) via athletic competence in older age, might permit less-masculine health care practices. Thus, when it came to the theme of healthy ageing, this exception signalled how for some, embracing the pleasures of (potentially) health harming practices offered protection from appearing over-concerned with their health. To neglect this would mean neglecting the wider contexts in which health-relevant decisions are made, amidst a plethora of competing priorities in later life.

Time for 'me' and the deterioration of 'we'
Another central theme identified through our categorical-content analysis as being an essential and fundamental element of participating in physical activity in later life was the value of interpersonal relationships. Through a concern with the experiences of non-elite older adults, this builds on observations regarding the nature and source of family influences on sport participation among midlife masters athletes by Dionigi et al. (2012). Primarily, our categorical-content analysis demonstrated the value of interpersonal relationships in facilitating and supporting physical activity involvement. Within this theme, physical activity was a shared hobby, or a pastime actively encouraged and supported by a partner (via information sharing, providing transport, providing emotional support). An exception here, was the presence of conflict within relationships resulting from people's engagment in an active lifestyle. While initial interpretations of these exceptions confirm previous research highlighting the value of physical activity in later life for providing individuals with 'me time' and respite from care giving roles (e.g. see Sims-Gould et al. 2010), closer scrutiny revealed the impact that regular participation in physical activity could have on the relationship per se between individuals. An example of this exception was noted in Mary's account of atttending T'ai Chi classes: I found just about two and a half years ago, I started thinking and I thought, you know, I've been a wife and a mother. I've done this all this time. A husband, a lack of confidence, I've lost a lot of my self-esteem. I've changed a lot getting married, changed a lot. Very much a dominant husband. And I thought, 'it's about time, I'm not and I've never been assertive and I thought, 'well, I'm going to have to try' . I'm not going to put up with it. It's caused a lot of trouble [between us] but I thought 'no, it's what I'm going to do for myself. I've just got to do it' . So we've had a very stormy two and a half years but, you know, every now and again I give in a bit and I think 'no, I'm not, I'm not going to. This is for me. It's important. I love physical activity and I'm going to do it. I've done enough for everybody else … I think he's accepted that. I said to him 'look, I love physical activity' . This is me being youthful again. I loved it as a child. This was me. I gave up so much getting married. Now this is "me time" again now. I'm being youthful. I'm doing the things that I enjoyed as a young person and this is what I want to do. And I think he doesn't like it and he thinks, he calls me selfish and all that. But he has, it's got a bit better. He's come to accept it a little more. First it was a nasty 'enjoy' as I went out the door and I would say 'yes, I will enjoy it' . (Mary,age 72) Exploiting this exception enabled insight into what interpersonal conflict might look and feel like within the context of later life physical activity. It reflects critical and controlling behaviours outlined by Roberts (2006) in her overview of destructive conflict domains. It also illustrates the impact that being physically active in older age may have for what Gergen (2009) describes as 'relational deterioration' . While we may strive for harmony in relationships (and indeed, our analysis demonstrated how physical activity in later life could be instrumental in this process), Gergen notes a series of conflicts, or counter-logics that can cause relationships to suffer.
A logic that had relevance for the exception we observed within this central theme was rationality versus its counter logic of emotionality. Elaborating on this concept, Gergen (2009) explains how relational deterioration can occur when the desire for individual action to be rooted in 'good reasons' (rationality) is countered by a prizing of emotions. Placing value on expressions of emotion (e.g. love, grief, compassion, frustration) to the extent that it becomes the main justification for action (e.g. participating in physical activity purely for the 'love' of it) can result in the action being discredited for its irrationality. For Gergen, these counter-logics of rationality and emotionality are pervasive within society, beckoning us all into degenerative relations. They are also highly gendered, with males typically being associated with rationality compared to women who often stereotyped as being irrational, and highly emotional.
Indeed, this exception that we observed within the central theme of 'the value of interpersonal relationships' resonates with Connell's (1987) early, but highly influential concept of hegemonic masculinity, which works to legitimise and maintain patriarchal relations. In particular, Mary's negative experiences of her husband in response to her desire to attend regular T'ai Chi classes highlights his complicity in sustaining the hierarchical gender order and the complex manner in which this intersects with her participation in active leisure (see also Henderson 2003;Im et al. 2008;McGannon and Busanich 2010). With respect to Mary's enduring identities as a mother and a wife, it also connects with feminist scholarship on the topic of care -a role predominantly undertaken by women (see Fine and Glendinning 2005). Despite growing evidence that engaging in physical activity can facilitate an (albeit temporary) escape from care roles, provide women with opportunities for self-care, and empowerment (Sims-Gould et al. 2010), care practices and responsibilities are more often negotiated in relation to the needs, desires, tastes, and individualities of the care receiver rather than the care giver (Maher et al. 2013).
Reinforcing the importance of focusing on exception data within central themes, Mary's story brings to the fore a potential (unintended) and less commonly articulated consequence of pursuing a physically active lifestyle in later life regarding the damage it can do to relationships in older age. Thus, going beyond the central theme in which the value of positive interpersonal relationships from physical activity participation is emphasised, to instead focusing on an exceptional story depicting interpersonal conflict, has been instructive for the final interpretations drawn from our data. It has sensitised us to the potential contribution that physical activity participation can make to relational deterioration by impacting negatively on spousal dynamics, communication patterns, boundaries, roles, patriarchal power relations, and notions of care. This progresses current conceptualisations toward more nuanced understandings of the various ways in which relationships (as opposed to merely individuals within a relationship) might also be impacted.

'Moving (on)' in social spaces and solitary places
A further central theme identified alluded to the value of physical activity for generating new, or indeed strengthening existing social networks following the death of a loved one (be it parent or spouse). Many participants explained how it was during the dark process of bereavement that a pre-arranged activity session gave them 'a reason to get out of bed' , interact with others in fun and sociable environments (Cooper and Thomas 2002) and experience feelings of being 'energised by different people' (Ziegler and Schwanen 2011, p. 758). In this way, being physically active in supportive social spaces was important for maintaining and improving wellbeing among older people by helping the bereaved to move through, and ultimately on from the grieving process (to the extent that this was ever entirely possible).
An exception to this pattern of engaging in physical activity spaces to encounter the company of others, was the role of physical activity in providing opportunities for solitude in meaningful places. This exception was exemplified by Colin: My mother died three years ago, and this is nothing to do with exercise really, but I started to miss my father who died forty years earlier and I suddenly found I was grieving for my father… I started grieving for my father and I went, so it was strange, I went and did St Oswald's way, which is a 97-mile-long distance trail in Northumberland, starts on Holy Island [North East of England]. But it goes through all the area that I was brought up in and spent my holidays in and I'd known with my father. So I did that in 6 days of walking and loved it … I realized then that I could do it (walking) and I needed to do it and, yeah you know, I thought I've got a few years left I really want to see what I can do. (Colin,age 69) The central theme emphasised a distraction from grief via the sociable encounters that involvement in physical activity afforded. In contrast, an exception within this theme emphasised the value of being solitary in active places to reflect upon and reconnect with loved ones lost. Indeed, as demonstrated in the example above, place was central here and our attention to this somewhat contradictory observation moves our understanding into largely unexplored territories regarding current knowledge of what it means to be physically active in later life. That is, the mechanisms by which physical activity can facilitate experiences of solitude in connection with meaningful places.
The importance of understanding how places become meaningful at specific moments in the life course is discussed by Smaldone et al. (2005), who advocate the conceptualisation of place as a process, rather than a static entity; much like, we would add, living a physically active lifestyle and indeed bereavement itself. Elaborating further, Smaldone et al. argue 'experiences that come with time create and strengthen emotional attachments to places, but places can become even more significant by helping to actually define a significant life stage for people' (p. 403). In the example above, St Oswald's Way came to define not only Colin's childhood through memories of family vacations and growing up, but also part of the transition into a new life stage; becoming an orphan. Experienced, of course, in many instances as almost a rite of passage into one's own (older) adulthood.
In addition to defining a significant life stage, meaning was attached to place (e.g. St Oswald's Way) and the solitary experiences it facilitated through movement (the act of walking). Indeed, while people will respond emotionally in different ways to different places, the same can be said for movement from place to place (Duff 2011). Thus, the exception of engaging in physical activity following bereavement to experience solitude illustrated how the grieving process wasn't progressed by merely being in a meaningful place, but by moving through it. In doing so, it supports recent theorising within health geography that advocates extending our theoretical understanding beyond static conceptions of "healthy places" to therapeutic mobilities. This takes into account the notion that movement itself can be conducive to wellbeing and health through both effective and affective means (Gatrell 2013).
Being attentive to this exception also advanced knowledge of relationality within the context of later life physical activity. It progressed understanding of this issue beyond the over-simplistic binary of exercising individually versus with others ('group based'). The exception did not reflect the more commonly recounted experiences of physical activity involving sociability and fun or feelings of being 'energised by different people' captured by the central theme (see Ziegler and Schwanen 2011, p. 758;Cooper and Thomas 2002), but nor did it resonate entirely with walkers experiences of sharing solitude with fellow walkers referred to by Sharpley and Jepson (2011). Rather, in this exception it was paradoxically, the absence (of a loved one) that ensured their very presence was felt as the bereaved moved through their memory filled, meaningful place.
This incorporation of absent others into the realm of walkable sociability offers an interesting and important extension of Doughty's (2013) emphasis on the social dynamics of embodied movement. Referring to the modalities of sharing therapeutic space, she notes how walking can support restorative experiences (such as bereavement) because of the imaginary connections and affective investments that are made by walking bodies, along with the possibilities for focusing on emotional issues as one uses the rhythms of walking to enter into a dialogue between landscape, body and mind. While Doughty's concern with sociability concerns 'walking with' other living people, her conceptualisation of mobile therapeutic space has relevance for these exceptional experiences of walking with imagined others, including those who have passed away or indeed past 'childhood' selves. In this way, building on Slavin's (2003, p. 11) account of pilgrims 'being with oneself while also walking beside others' , the exception alludes to the experience of spiritual sociability produced by feelings of proximity to imagined others while moving through a meaningful place.
By directing analytical attention toward the importance of place, therapeutic mobilities and spiritual sociabilities, this exception to the central theme provided 'a more flexible social politics of interaction' (Doughty 2013, p. 145) relative to later life physical activity. It thereby advanced current knowledge regarding the intersection between bereavement and physical activity in older age by extending it beyond the established notion that being or becoming active can enhance social networks, which in turn may reduce feelings of social isolation and support people through the grieving process.

Discussion
In this paper, we have presented three examples of how exception (or 'outlier') data was incorporated into our analyses and subsequently advanced our knowledge of later life physical activity beyond 'more of the same' . For example, rather than conclude with the somewhat overly simplistic assertion that 'physical activity is an important component of healthy ageing' , a focus on exceptional data illuminated how, in some instances, later life physical activity can also form part of an ambivalent health strategy, which seeks to integrate desire, pleasure and enjoyment across a range of health practices (enhancing and harming). This observation builds on a currently small but significant collection of work seeking to amplify and theorise the lesser heard affectual accounts of living in, and through, an ageing body (e.g. Milligan et al. 2005; Orr 2014, see also Ross and Mirowsky 2008). Similarly, rather than merely pronouncing the value of interpersonal relationships in facilitating and supporting physical activity involvement, a focus on exceptions signalled the manner in which physically active lifestyles might be initiated, reignited, developed and maintained throughout older age amidst (sometimes 'stormy') interactions and (patriarchal) gendered relationships. Finally, rather than simply reinforce the value of physical activity for generating new, or indeed strengthening existing social networks following the death of a loved one, incorporating exceptions into our analyses resulted in us venturing beyond relationships with those who are physically encountered, to consider relationships with those who are imagined (e.g. the deceased). The way in which these relationships with (imagined) others intersect with meaningful places via movement extends existing knowledge regarding the way in which therapeutic mechanisms might shape experiences of physical activity during this life stage. Accordingly, following McPherson and Horne (2006), we have argued that being attentive to exceptions within qualitative data has the capacity to develop methodological expertise and as a consequence, the rigour and credibility of research findings. Exceptional stories are not the result of unfortunate anomalies in our design decisions, nor poor interviewing technique as might be the case in Connelly and Peltzer's (2016) description of 'underdeveloped themes' . Rather, they represent plausible manifestations of important human diversities that differ from the dominant narratives within the data and contradict our more immediate theoretical and thematic interpretations. Had we not pursued the exceptions in our own data, our confidence in the explanatory value of our research, specifically in relation to the central themes of healthy ageing, relationships, and death and bereavement, would have been undermined.
All of this, however, is not to negate the worth of identifying central themes and storylines within qualitative data for organising and describing a data-set in rich detail, nor facilitating the interpretation of various aspects of the research topic (Braun and Clarke 2006). Rather, it is to remain mindful, as indeed noted by Braun and Clarke, that identifying central themes is a core skill within qualitative research that is useful for conducting many other forms of analysis. Thus, our intention is to urge qualitative researchers in sport, exercise and health to avoid discounting diversity, anomalies or exceptions in the pursuit of prevalence, patterning and commonality. There are instances where identifying central themes and storylines could be the first rather than final step of data analysis. Research concerned with older adults' experiences of sport and physical activity is, we believe, one such instance.
That noted, amassing rich and varied interpretations of a given subject, does not come without further challenges, not least disciplinary differences and also presentational requirements of some journals. These challenges also relate to the translation of research into policy and practice. Discussing the different cultures of policy-making and qualitative research, Donmoyer (2012) observes that government funders are essentially interested in knowing 'what works' . Accordingly, they require researchers to provide answers to their questions in ways that are not littered with detail nor obscured by endless clarification. In short, Donmoyer explains, the role of the policy-maker is to remove complexity. 'Too much complexity -a focus on too many "on the other hands,' too much contextual variation -invites immobilisation" (p. 801). Yet this impatience toward complexity from the worlds of policy and practice and the type of research that might be encouraged as a result has caused disquiet in some corners of the qualitative community. Here, critiques have been levied at the 'simplistic and mechanistic approaches' that equate data reduction and coding as rigorous analysis, used to produce knowledge devoid of critical reflection and contextual considerations (Koro-Ljungberg and Mazzei 2012, p. 728). The 'politics of simplification' -largely driven by policy and funding agencies -are problematic, argue Koro-Ljunberg and Mazzai, not least because they are used to promote scientific reductionism, uninformed scholarship, atheoretical interpretation and a diminishment of scholarly rigour (p. 728).
Elsewhere, scholars have bemoaned the limited and somewhat unimaginative ways that policy currently mis or selectively use research, leading to a greater prominence of limited and unimaginative, yet (deemed to be) 'impactful' research, being produced by commissioners. For example, Green (2009) has critiqued public health policy for its over-reliance on intervention based walking studies, which have led to walking being encouraged 'as if it were solely an embodied activity, to do with using our physical bipedalism in an asocial and non-material way' (p. 35). Extending this further, Gatrell (2013) asserts that this traditional public health, or health promotion strategy overlooks the embeddedness of walking; its environmental, moral and political backcloth. In a similar vein, Phoenix and Orr (2014) have noted how a focus on 'the usual suspects' (i.e. the physical health outcomes) throughout much of the discourse surrounding the promotion of physical activity in older age, is at odds with the various pleasures (documented, sensual, habitual, immersive) that many older adults describe as being central to their experiences of being active. What's needed, they argue, is a more sustained engagement with the complexities inherent to lifelong participation in physical activity and an emphasis on its affectual nature.
In addition to the disjuncture between policy/programmes and individual meaning/experience that can potentially arise from the simplistic and mechanistic approaches to research referred to above, ethical issues are at stake when one runs rough shod over the complexity that is inherent to growing older. For example, in relation to the current paper, our concern with data complexity via a focus on exceptions highlights the need for a greater sensitivity toward the unintended consequences of uncritically advocating older adults getting out and getting active (with others). Without due consideration of the implications this could have for other spheres of wellbeing, alternative but important mechanisms for protecting one's health and wellbeing might be inadvertently harmed. 3 Physical activity (and other such health behaviours) are but one route to optimising health in older age. They should not, therefore, be promoted as a panacea for 'ageing well' , ignorant of the impact these routes might have on other equally important means of navigating one's journey into older age (e.g. spousal relationships, friendships forged from time spent in the pub). Thus, in advocating the incorporation of exceptional data into ones analyses, we are mindful of Frank's (2010, p. 72-73) reference to Foucault's 'practice of criticism' as he urges social researchers to develop the skills and approach of an 'intellectual craftsman' . This, he asserts, involves engaging in movement of thought that develops, utilises and revises conceptual tools that are embedded within relationships and social practices, thereby enabling a critically informed interpretation of the data.
With these points in mind, despite the additional challenges that too many 'on the other hands' , or problematised findings might pose for the translation of qualitative knowledge to policy and practice (a challenge that we believe qualitative researchers must take some responsibility for developing the communication skills to overcome 4 ), we remain convinced of the need for complexity to continue being identified, understood and communicated -both in and outside of academe. This is especially the case in the field of ageing and physical activity, which as argued earlier in this paper, is at risk of falling foul to simplistic, mechanistic inquiry and subsequently stagnating. Maintaining a commitment to complexity requires engagement with methodological multiplicity, which in turn, can orientate researchers toward conceptual, analytical, and interpretive spaces that can meet the needs of ever-changing communities of practice (Koro-Ljunberg 2012). It also calls for what what Koro-Ljungberg and Mazzei (2012, p. 730) refer to as 'deliberate and slow methods' , which assist qualitative researchers engage in activities that avoid automation, prediction and habituated assumptions that restrict types of knowledge produced. Deliberate and slow methods, they propose, 'can be seen as a way to work between theory and research in social, moral and reflective ways that interrogate contradictions, differences and dissent' . It is within the context of deliberate, slow, methodological multiplicity that exceptions within qualitative data have an important role to play in progressing theory, methods and policy. Notes 1. Funding information temporarily withheld to ensure anonymity. 2. Following Ali Sayed and French (2016), males are defined as heavy drinkers if they have consumed, on average, more than 14 drinks (14 units) per week during the past year. 3. Our point here is not to support marital relationships that are experienced as controlling and prevent a partner from pursuing hobbies such as physical activity (as implied in our own data), but rather to signal broader issues regarding the significance of companionship, and sense of intimacy and closeness that can be experienced via caring for another. Both of which, in certain circumstances, can be negatively impacted by the pursuit of a physically active lifestyle, particularly when relationality is not considered. 4. Although beyond the scope of the current paper, readers are directed to Donmoyer (2012) for practical strategies of how qualitative researchers might influence policymakers and the policy-making process while staying true to what this author refers to as one's 'methodological soul' (p. 798).

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Economic and Social Research Council [ES/M001709/2, ES/I009779/1].

Notes on contributors
Cassandra Phoenix is an Associate Professor in the Department for Health at the University of Bath. Her current research uses narrative approaches and visual methods to understand what it means to 'age well' in a variety of contexts across the life course. Her work is underpinned by an ongoing commitment to the use and development of qualitative methods to address complex health issues. Cassandra has published widely in health, method, and gerontology journals. Her research has been funded by a range of research councils and charities.
Noreen Orr is a Research Fellow NIHR CLAHRC South West (PenCLAHRC) in the University of Exeter Medical School, University of Exeter, UK. She is an experienced qualitative researcher and her research focuses on understanding older people's experiences of ageing well.